Factors Affecting the Distribution ofAtyid Shrimps inTwo Tropical Insular Rivers l

Trina Leberer-,3 and Stephen G. Nelson2,4

Abstract: We investigated factors affecting distribution of atyid shrimps, com­mon inhabitants of insular freshwater ecosystems. Several abiotic and bioticvariables were measured to determine their influence on atyid shrimp densitiesin two streams on the western Pacific island of Guam. Randomly selected sites,composed of three habitat types (rimes, runs, and pools), were surveyed in therainy and dry seasons. We made visual counts of instream fauna in 2-m2 quad­rats within each site. Various statistical analyses suggested that habitat type is amajor factor affecting atyid distribution on Guam. However, results of a trans­plant experiment, conducted to test the effect of predators on atyid distributiondirectly, were noteworthy: no atyids remained in pools containing the trans­planted jungle perch Kuhlia rupestris in the field. Our data indicate that bothenvironmental factors and faunal interactions may be important influences onatyid distribution.

ATYID SHRIMPS ARE common in the fresh­water habitats of oceanic islands. There areseven recorded species of stream atyids on thewestern Pacific island of Guam: Atyoida pilipes(Newport, 1847), Atyopsis spinipes (Newport,1847), Caridina longirostris (Milne-Edwards,1837), Caridina nilotica (Roux, 1833), Caridinaserratirostris de Man, 1892, Caridina typusMilne-Edwards, 1837, and Caridina weberi deMan, 1892. All except Atyopsis spinipes wereseen during this study. Caridina longirostrisand C. weberi are new records for Guam (thisstudy). All of these species are native toGuam, but none is endeInic. Like most otheratyids (Shokita 1976, Benzie 1982), all thoseoccurring in the rivers of Guam are am­phidromous, with their larvae drifting down­stream to the ocean, developing in the

1 Manuscript accepted 5 January 2001.2 University of Guam Marine Laboratory, UOG

Station, Mangilao, Guam 96923.3 Current address: Department of Agriculture, Divi­

sion of Aquatic and Wildlife Resources, 192" Dairy Road,Mangilao, Guam 96923.

4 Current address: Environmental Research Labora­tory, University of Arizona, 2601 E. Airport Drive, Tuc­son International Airport, Tucson, Arizona 85706-6985.

Pacific Science (2001), vol. 55, no. 4:389-398© 2001 by University of Hawai'i PressAll rights reserved

plankton, and Inigrating back upstream as ju­veniles. Although some previous studies haveaddressed the postlarval ecology of atyids invarious geographical locations (Hart 1981,Carpenter 1983, Dudgeon 1985, de Silva1988, Crowl and Covich 1994), less infor­mation is available from the tropical PacificRegion. Couret (1976) provided a compre­hensive summary of the biology of the Ha­waiian atyid Atyoida bisulcata Randall, 1840,including aspects of its ecology. Bright (1979)included some ecological information in hisstudy of freshwater ecosystems in Palau. Ellis­Neill (1987) concluded that current velocityand a greater abundance of leaf litter are im­portant influences of invertebrate distribution(including atyids) on Guam. The ecology ofatyids on Yap was briefly mentioned in Bright(1989).

In this study, we addressed the followingquestion: Which environmental and faunalfactors affect the distribution of atyids onGuam? We analyzed several abiotic and bi­otic variables, including season, reach, habi­tat, pH, water temperature, maximum depth,canopy cover, substrate composition, abun­dance of aquatic vegetation, current velocity,and densities of other riverine fauna. Prelim­inary observations suggested that predationwas an important faunal factor influencingatyid distribution. We examined predation in

389

390

two ways. First, we compared densities ofatyid species with potential predator densitiesin a correlation analysis. In the streams ofGuam, potential predators include the jungleperch Kuhlia rupestris (Lacepede, 1802); thefreshwater eel Anguilla marmorata Quoy &Gaimard, 1824; the sleeper goby Eleotris fusca(Bloch & Schneider, 1801); the Mariana gobyAwaous guamensis (euvier & Valenciennes,1837); and the Tahitian prawn Macrobrachiumlar (Fabricius, 1798). Stomach contents ofthese species (except M. lar) have been foundto contain atyids (R. B. Tibbatts, pers.comm.). Second, to test the effect of preda­tors on atyid distribution directly, we con­ducted a transplant experiment in the field,involving K rupestris. Kuhlia are free-swim­ming, visual predators, occurring only in thelower reaches of rivers, below barrier water­falls. If predation by Kuhlia is an importantinfluence on the distribution of atyids in thestreams of Guam, then densities of atyid spe­cies would be expected to be lower in poolscontaining transplanted K rupestris than incontrol pools containing no K rupestris.

MATERIALS AND METHODS

Study Area

We collected data from two rivers (Figure 1).The Asmafines River (coordinates at themouth: lat. 13° 19' 37" N, long. 144° 39' 03"E) is a high-gradient stream located in thesouth on Guam's steeper western slope, witha perennial channel length of 1341 m and anelevation of 134 m at the headwaters (Bestand Davidson 1981). The Ugum River (co­ordinates at the confluence: lat. 13° 20' 11"N, long. 144° 45' 08" E) is a long, low­gradient stream located in one of the largedrainage basins characteristic of the south­east. It is a major tributary of the TalofofoRiver and has a main channel length of11,460 m and an elevation of 183 m at theheadwaters (Best and Davidson 1981). Bothrivers are considered relatively undisturbed,but the lower reaches of the Ugum are ob­structed by a weir. The weir was built in 1992by the Public Utility Agency of Guam and isestimated to supply between 7 and 11 millionliters of water per day to villages in the

PACIFIC SCIENCE· October 2001

southeast (Wiles and Ritter 1993). The effectof this water removal on atyid densities anddistribution is poorly understood.

Study Sites and Field Methodology

We selected sites with a modified version ofthe stratified random sampling method ofBaker and Foster (1992). Habitat types wereclassified qualitatively as rimes (areas withcurrent, surface turbulence, and emergentrocks), runs (areas with current, no surfaceturbulence, and no emergent rocks), andpools (areas with little or no current and nosurface turbulence). Three reaches (upper,middle, and lower) were identified in bothrivers. The upper reaches were characterizedby a steep gradient, short drops, and plungepools with a rock or boulder substrate, andflanked by dense riparian vegetation. Themiddle reaches were characterized by a lowto medium gradient, substrate dominated byrocks and boulders with some sandy areas,and thinning riparian vegetation. The lowerreaches were characterized by slow currents,wide and deep water, and a dominant sub­strate of sand and/or mud. We intended tochoose six sites within each reach (two of eachhabitat type). However, we did not encounterany pools in the lower reach of the Ugum,and there was a paucity of rimes in both thelower reach of the Ugum and the upper reachof the Asmafines. In addition, during the pe­riod of high flow, one run on the lower reachof the U gum was impossible to survey. Wesurveyed extra rimes and extra runs (one inplace of the missing pool) in the middle andupper reaches of the Ugum. Therefore, wesurveyed five rimes, six runs, and six pools onthe Asmafines River and seven rimes, sevenruns, and five pools on the Ugum River, for atotal of 36 sites. Because previous observa­tions indicated that current velocity mighthave a major effect on atyid distribution(Ellis-Neill 1987), data were collected fromthe same quadrats (when possible) duringperiods of low flow (March 1996-May 1996)and high flow (September 1996-January1997).

We chose sites on both rivers with theaid of a random numbers table. Numbers

10 km

Philippine Sea

Asmafines River

FIGURE 1. Map of Guam indicating geographic location of study areas.

Pacific Ocean

N

W~Es

" " " " " " " " " " " " " "

Ugum River

392

selected represented the number of habitattypes upstream from access points to thereach. For example, for pools, selection of 3and 5 indicated that we surveyed the thirdand fifth pools encountered upstream fromthe point of access to the reach. Sets of num­bers for each type of habitat were selected.

We conducted visual counts in one virtual2-m2 (1.4 by 1.4 m) quadrat within each site.We measured the parameters of each quadratwith a tape measure before we began thecounts. This quadrat size represented at least10% of each site. Because adults of all atyidspecies are found more frequently near theedges of rivers, we positioned quadrats alongthe bank, arbitrarily choosing the right sideof the river (when facing upstream) and theupstream end of the habitat type. We ap­proached the quadrat from downstream so asnot to disturb the shrimp. We performed vi­sual counts while snorkeling in the stream orfrom the bank. Although adults of Atyoida pi­Iipes can be easily identified in the field, adultsof the five species of Caridina resemble oneanother (Bouvier 1925) and cannot be dis­tinguished without the aid of a microscope.Thus, during visual counts, atyids were iden­tified to genus. Upon completion of visualcounts, we collected a subsample of atyids(with a dip net) from the quadrat. We fixedindividuals in 80% ethanol for later identifi­cation.

In addition to atyid density, the followingvariables were evaluated, in each quadrat, ateach site: (1) pH; (2) water temperature; (3)maximum depth in quadrat; (4) percentagevegetation cover; (5) substrate type (percent­age bedrock; percentage boulders; percentagerubble; percentage gravel; percentage sand;percentage silt); (6) abundance of aquaticvegetation; (7) current velocity; (8) density ofKuhlia rupestris; (9) density of Anguilla mar­morata; (10) density of Awaous guamensis; (11)density of Eleotris fusca; (12) density of Mac­robrachium lar; and (13) density of all othergobies combined, including Sicyopterus macro­stetholepis (Bleeker, 1853), Sicyopus sp., Sten­ogobius sp., Stiphodon percnopterygionus Watson& Chen, 1998, and Stiphodon sp.

We measured pH with a pH meter (SperScientific Digital 840008) and temperature

PACIFIC SCIENCE· October 2001

with a mercury thermometer, both on site.The canopy cover directly over the quadratwas estimated visually to the nearest 10%.We tested this method against the resultsof a forest densiometer (Robert E. LemmonModel-C) and found the visual estimate to beaccurate and more expedient. The percentagecover of each substratum type was also esti­mated visually. We categorized substrata asbedrock, boulders (>256 mm; "head sized"),rubble (>64-256 mm; "fist sized"), gravel(>2-64 mm; "thumb sized"), sand (>0.2-2mm; "grain sized"), and silt (material 2 Jlm­0.2 mm that can be suspended in the watercolumn) (modified from Helm et al. 1985).The percentage cover of each substratumtype was estimated in one-fourth incrementsof the quadrat and these increments wereassigned the following values: 0, absent;1, 25%; 2, 25-50%; 3, 50-75%; and 4, 75­100%. We categorized the aquatic vegetationqualitatively as 0, absent; 1, present (found in<50% of quadrat); or 2, abundant (found in>50% of quadrat). Current velocity wasmeasured at the center of the quadrat, at0.6 x the maximum depth, with a velocitymeter (Marsh-McBimey MMI Model 2000Flo-mate). We determined population den­sities of other species (variables 8-13) withvisual counts, at the same time and in thesame way as atyid densities.

Transplant Experiment

In addition to examining the data with variousstatistical analyses, we conducted a transplantexperiment to test the effect of predationdirectly. We identified three pairs of pools,each containing comparable atyid densities(determined by separate visual surveys, as de­scribed earlier, but of the entire pool), abovethe barrier waterfall on the Asmafines River.The upstream and downstream access pointsof these pools were fenced with 3-cm meshwire. This mesh size was sufficient to containthe Kuhlia. We did not try to contain theatyids within the pools, because the requiredmesh size impeded water flow. Containmentof atyids is also made difficult by their abilityto migrate across moist terrestrial substrate(Jalihal et al. 1994). We then measured atyid

Factors Affecting the Distribution of Atyid Shrimps Leberer and Nelson 393

densities again, using visual surveys, after thescreens were in place. Kuhlia were capturedfrom the lower reaches of the AsmafinesRiver, transferred to the pools in 20-literbuckets, and released. We placed one or twoKuhlia in each of three randomly selectedpools (one from each pair). One each wasplaced in two pools (area = 5.22 m2, densityof Kuhlia = 0.2 per square meter; area = 7.30m2

, density of Kuhlia = 0.1 per square meter).We placed two Kuhlia in another pool(area = 17.55 m2

, density of Kuhlia = 0.1 persquare meter). These densities fell within therange of natural densities of Kuhlia deter­mined in this study (0.0-5.0 per square me­ter). After 24 hr (the pool with two Kuhlia)and 48 hr (the other two pools), we used vi­sual surveys to determine atyid densities againfor all pools.

Data Analyses

We performed data analyses with the Stat­view Statistical System for Macintosh (AbacusConcepts 1995). For all parametric tests, weconducted Levene's test for homogeneity ofthe variances and performed appropriatetransformations as needed. When data didnot meet the assumptions of analysis of vari­ance (ANOVA), even after the transforma­tions, appropriate nonparametric tests wereconducted (Sokal and Rohlf 1995).

RESULTS

Environmental Factors

Figure 2 shows the distribution of atyid spe­cies in each river. The densities used in Fig­ure 2 are based on the subsamples collectedfrom each quadrat and are not comprehen­sive. Different species of Caridina appear topredominate in each river. Caridina typus andC. weberi occurred in higher densities in ourcollections from the Asmafines River, where­as C. nilotica was the dominant species in col­lections from the Ugum River. Densities ofAtyoida (determined by visual counts) aresignificantly lower in the Asmafines River(Mann-Whitney U, tied Z = -2.76, n = 34,tied P = 0.006), but comparable densities of

9

8II Asmafines

" 7::E Ugum~ 6:l0' 5rnII 4.,.~

OJ 3.g.S:

2'6oS 1

0.., ..,

".;::

" " .~ .~

~ ~.~ ~<; ::: :::

" '" ~ .~ ~

~ " " " }:§~ " ~c ... :§

Q (J ... "'<: d (J ..,"" "

~:§...... (J(J

FIGURE 2. Distribution of six species of atyid shrimp inthe Asmafines and Ugum Rivers of Guam. Mean den­sities were determined from collections of specimenswithin O.25-m2 areas with a dip net. Capped bars repre­sent standard errors of the means.

Caridina (determined by visual counts) oc­curred in both rivers (Mann-Whitney U, tiedZ = -0.703, n = 34, tied P = 0.48).

We compared each environmental variablewith densities of Atyoida and densities ofCaridina (Table 1). Although many relation­ships were not found to be significant, somevariables appear to be important influences ofatyid distribution. Densities of Atyoida werepositively correlated with temperature andaquatic vegetation and negatively correlatedwith depth (Table 1). Densities of Caridinawere positively correlated with depth andnegatively correlated with pH (Table 1). Inaddition, densities of Atyoida were sig­nificantly higher in rimes than in runs andpools (Kruskal-Wallis, H [corrected forties] = 6.84, n = 34, tied P = 0.03). Con­versely, densities of Caridina were signifi­cantly higher in runs and pools than in rimes(Kruskal-Wallis, H [corrected for ties] =8.57, n = 34, tied P = 0.01).

Faunal Interactions

The effects of faunal interactions on atyiddensities and distribution were also analyzed.

394 PACIFIC SCIENCE· October 2001

TABLE 1

Correlation Z Tests for Densities of Atyid Genera and Environmental Variables

Variables

Canopy coverpHTemperatureDepthSubstrateAquatic vegetationCurrent velocity

* Significant.

Atyoida

Z = -0.084, n = 72, P = 0.93Z = -0.463, n = 72, P = 0.64Z = 2.399, n = 72, P = 0.017*Z = -2.453, n = 72, P = 0.014*Z = 0.089, n = 72, P = 0.93Z = 2.440, n = 72, P = 0.015*Z = 1.223, n = 72, P = 0.22

Caridina

Z = -0.828, n = 72, P = 0.41Z = -3.483, n = 72, P = 0.0005*Z = -0.708, n = 72, P = 0.48Z = 3.31, n = 72, P = 0.0009*Z = 1.331, n = 72, P = 0.18Z = 0.001, n = 72, P = 0.9994Z = -1.042, n = 72, P = 0.30

60 ,-;::- 50B"e 40

.£~~ " 30"<:r 0Q:~g, 20 0.:s~

";;: ~<'0:9 10 §OU.:::

"0e- O E7{ffi() 0 0 -e-

-10 +-,------j~-,-~,---.-,~,--,....,~-,-~,---.-,

-5 0 5 10 15 20 25 30 35 40Atyoida Density

(individuals per square meter)

FIGURE 3. Scatterplot of density ofAtyoida versus densityof Caridina. Densities were determined by visual counts.

Densities of the two atyid genera are nega­tively correlated, but not significantly so(Correlation Z test, Z = -0.94, n = 72, P =0.35). The two genera do not frequently co­occur (Figure 3), which corresponds with

their preponderance in different habitats.Densities of atyids were not significantlynegatively correlated with predator densities,neither separately nor when all potentialpredators were combined (Table 2). How­ever, we did not see adult atyids where Kuhliaoccur, in the lower reaches, below the barrierwaterfalls. We found only juvenile atyids,buried within gravelly substrata, in habitatscontaining Kuhlia.

Measurements of the carapace lengths(from the postorbital edge to the posteriorend of the carapace, measured at the midline)of the preserved specimens were taken tocompare whether shrimp size distribution re­flected atyid amphidromy or the presence ofKahlia. Mean shrimp carapace length is sig­nificantly larger above the barrier waterfallsthan below in both rivers and for all speciescombined (Kruskal-Wallis, H [corrected forties] = 86.92, n = 171, tied P = 0.0001). Inaddition, we compared the carapace lengths

TABLE 2

Correlation Z Tests for Densities of Atyid Genera and Predator Densities

Variables

Kuhlia rupestrisAnguilla mamzorataAwaous guamensisEleotris fuscaMacrobrachium larAll predators combined

* Significant.

Atyoida

Z = -1.243, n = 72, P = 0.21Z = -0.612, n = 72, P = 0.54Z = -0.738, n = 72, P = 0.46Z = -0.612, n = 72, P = 0.54Z = -0.392, n = 72, P = 0.70Z = -0.949, n = 72, P = 0.34

Caridina

Z = -1.023, n = 72, P = 0.31Z = 0.008, n = 72, P = 0.99Z = 0.045, n = 72, P = 0.96Z = -0.312, n = 72, P = 0.75Z = 0.759, n = 72, P = 0.45Z = 0.455, n = 72, P = 0.65

Factors Affecting the Distribution of Atyid Shrimps . Leberer and Nelson 395

(for all species combined) from sites nearestthe top of the barrier waterfalls on each riverwith those from sites farther upstream. Therewas no significant difference between sites(Mann-Whitney U, tied Z = -0.356, n =129, tied P = 0.72).

Transplant Experiment

We conducted a transplant experiment in­volving Kuhlia to test directly their impor­tance as predators on atyids. Before wetransplanted Kuhlia, all experimental andcontrol pools contained atyids at densitiesfrom 0.92 to 1.32 per square meter. We sawno atyids in pools containing transplantedKuhlia. Conversely, atyid densities in controlpools did not differ significantly during theexperiment (paired t-test, t = -1.999, df = 2,P = 0.184). After the experiment, barrierswere removed and after a few months ofheavy rains, the experimental pools werechecked again. No Kuhlia remained in thesepools, and natural densities of atyid shrimphad returned.

DISCUSSION

We found Atyoida and Caridina in differenthabitats. The two genera were also correlatedwith different environmental variables (Table1). Atyoida were seen mainly in rimes, whichcontain shallower water. This correspondswith their negative correlation with depth.They were positively correlated with bothtemperature and aquatic vegetation, whichwere both significantly higher in the dryseason than during the rainy season (one-wayANOVA, F = 21.32, df = 70, P = 0.0001;one-way ANOVA, F = 7.75, df = 70, P =0.007, respectively). Densities ofAtyoida werealso higher in the dry season, but not sig­nificantly so (Mann-Whitney U, tied Z =-1.524, n = 72, tied P = 0.13). Caridina werefound more often in runs and pools, both as­sociated with deeper water. This is supportedby their positive correlation with depth. Theywere also negatively correlated with pH,which was significantly higher in the Asma­fines River, due to the presence of a lime­stone cap located above the river (one-way

ANOVA, F = 50.92, df = 70, P = 0.0001).Conversely, densities of Caridina were lowerin the Asmafines, but not significantly so(Mann-Whitney U, tied Z = -0.703, n = 34,tied P = 0.48).

The association of Atyoida with riffle areasand Caridina with runs and pools is similar tothe findings of Chace (1983) and Ellis-Neill(1987), but contrasts with those of Bright(1989). This habitat differentiation is unlikelyto be the result of either interference or ex­ploitation competition. The two genera arenot obviously aggressive toward each otherin places where they do coexist (pers. obs.).They are both detritivores, and thus theirfood supply is probably not limited (Couret1976). Their predominance in different hab­itats corresponds with morphological adapta­tions to these habitats. Atyoida possess longersetae appropriate for filter feeding in highervelocity water (Couret 1976, Chace 1983,Ellis-Neill 1987), whereas Caridina have bothshort and long setae appropriate for filteringor scraping in a wider range of habitats(Bouvier 1925, Ellis-Neill 1987).

We saw no negative correlation betweenatyid and individual or combined predatordensities. However, two known predators,Ang;uilla marmorata and Eleotris fusca, werealmost certainly underrepresented in thevisual surveys. Ang;uilla marmorata is moreactive at night and E. fusca is highly cryptic,found most frequently under cover, and spe­cializes in ambushing its prey. On GuamAwaous g;uamensis is most often seen feedingin areas with soft substrates, and, althoughatyids have been found in their stomach, theirgut contents mainly comprise algae, sand, andoccasionally, other invertebrates (R. B. Tib­batts, pers. comm.). Moreover, all of thesepredatory fishes are present throughout theriver, co-occurring with atyids. Wellborn andRobinson (1991), through predator exclusionexperiments in a Texas reservoir, concludedthat predation by fish does not substantiallyaffect abundances of macroarthropods thatnaturally co-occur with them. They sug­gested that these macroarthropods possessantipredator defenses. Atyids may escapepredation by A. marmorata, E. fusca, and A.g;uamensis to a certain extent by reducing

396

activity at night (to avoid A. marmorata) andby occupying relatively exposed, hard sub­strata during the day (to avoid E. fusca andA. guamensis). The prawn Macrobrachium laralso occurs with both genera of atyids in highdensities and has not been seen to consumethem, neither in the field nor in laboratoryaquariums (pers. obs.). This contrasts withthe findings of Couret (1976) and Crowland Covich (1994), which demonstrated thatprawns in the genus Macrobrachium preyupon atyids in Hawai'i and Puerto Rico,respectively. However, in Hawai'i, Macro­brachium lar is an introduced species.

Predation by Kuhlia appears to be a factorinfluencing atyid distribution. The jungleperch Kuhlia rupestris is the only predator thatdoes not co-occur with adult atyids, and itwas shown to have a considerable effect onatyids in our transplant experiment. This dif­fers with the results of exclusion/inclusionexperiments in temperate areas. For example,Allan (1982) found that a reduction in troutdensities in streams in Colorado did notresult in a significant increase in densitiesof benthic invertebrates. Similarly, Reice andEdwards (1986) conducted both exclusionand inclusion experiments in two Canadianstreams (one that naturally contains fish andone that does not) and concluded that brooktrout do not have a major effect on the dis­tribution of benthic invertebrate communitiesin those streams. Gilinsky (1984) reportedmixed results from exclusion and inclusionexperiments with bluegill sunfish in a NorthCarolina pond. She found that the effectof fish predation on benthic macroinverte­brates was dependent on season and habitatcomplexity. However, Fraser et al. (1995)performed both inclusion and exclusion ex­periments in streams on the tropical island ofTrinidad and found that in areas split bybarrier waterfalls, predators can produce dis­jointed prey distributional patterns, by bothconsumption and by causing prey to ascendcascades. In addition, Reznick et al. (1997)moved guppies from pools with many preda­tors below barrier waterfalls into pools withonly one predator above waterfalls in Trini­dadian rivers. The guppies exposed to less

PACIFIC SCIENCE· October 2001

predation matured at a larger size at a laterage and produced fewer, larger young.

The results of the transplant experiment inour study suggest either a behavioral responseby the atyids, such as hiding or migrating toadjacent areas, or consumption by Kuhlia, orboth. Analysis of stomach contents of recap­tured Kuhlia would have been helpful in con­firming the latter, but we could not recaptureKuhlia with methods that are nondestructiveto the stream ecosystem. Laboratory experi­ments involving Kithlia and atyids mightprove useful in determining to what extentshrimp can hide or escape and to what extentthey are eaten.

The influence of Kuhlia on atyid distribu­tion is also supported by the difference inshrimp carapace length from below and abovethe waterfalls. We found that only small atyidsburied within substrata occur with Kuhlia. Ifdifferences in carapace length reflect the am­phidromous lifestyle of atyids instead of theinfluence of Kuhlia, a gradual increase insize might be expected as distance from themouth of the river increases. However, cara­pace length was not significantly larger insites farther from the top of the waterfallsthan in sites closest to the top of the water­falls. This suggests that Kuhlia are an influ­ence on atyid size distribution.

Although habitat characteristics appear tobe a major factor affecting the distribution ofatyid shrimps in the streams of Guam, ourtransplant experiment also illustrated the im­portance of predation by the jungle perchKuhlia rupestris on atyid distribution.

ACKNOWLEDGMENTS

We thank Frank Camacho, Michelle Gaither,and Brent Tibbatts for their assistance in thefield and for their much-appreciated senses ofhumor. We wish also to express gratitude tothe faculty, staff, and graduate students at theUniversity of Guam Marine Laboratory. Wethank Rosemary Leberer for locating refer­ences unavailable to us. Robert Rowan pro­vided valuable editorial comments on themanuscript. Finally, we thank Greg Fulcherfor his support.

Factors Affecting the Distribution of Atyid Shrimps . Leberer and Nelson 397

Literature Cited

Abacus Concepts. 1995. Statview StatisticalSystem for Macintosh. PowerPC version.Abacus Concepts, Berkeley, California.

Allan, J. D. 1982. The effects of reduction introut density on the invertebrate com­munity of a mountain stream. Ecology63:1444-1455.

Baker, J. A, and S. A Foster. 1992. Estimat­ing density and abundance of gobies inHawaiian streams. Rep. Hawaii Div.Aquat. Res.

Benzie, J. A H. 1982. The complete larvaldevelopment of Caridina mccullochi Roux,1926 (Decapoda, Atyidae) reared in thelaboratory. J. Crustacean BioI. 2:493-513.

Best, B. R, and C. E. Davidson. 1981. In­ventory and atlas of the inland aquaticecosystems of the Marianas Archipelago.Univ. Guam Mar. Lab. Tech. Rep. 75.

Bouvier, E.-L. 1925. Recherches sur la mor­phologie, les variations, la distributiongeographique des crevettes de la familledes Atyides. Eneyc1. Entomo1. 4:1-370.

Bright, G. R 1979. The inland waters ofPalau, Caroline Islands. Office of the ChiefConservationist, Trust Territory of thePacific Islands, Palau.

---. 1989. The freshwater decapod crus­taceans of Yap, Caroline Islands. Pages33-36 in S. G. Nelson, ed. The inlandaquatic habitats of Yap. University ofGuam Marine Laboratory, Mangilao,Guam.

Carpenter, A. 1983. Population biology ofthe freshwater shrimp Paratya curvirostris(Heller, 1862) (Decapoda: Atyidae). N. Z.J. Mar. Freshwater Res. 17:147-158.

Chace, F. A., Jr. 1983. The Atya-like shrimpsof the Indo-Pacific region (Decapoda:Atyidae). Smithson. Contrib. Zoo1. 384.

Couret, C. L., Jr. 1976. The biology andtaxonomy of a freshwater shrimp, Atyabisukata Randall, endemic to the HawaiianIslands. M.S. thesis, University of Hawai'iat Manoa, Honolulu.

Crowl, T A, and A P. Covich. 1994. Re­sponses of a freshwater shrimp to chemicaland tactile stimuli from a large decapod

predator. J. North Am. Bentho1. Soc.13:291-298.

De Silva, K. H. G. M. 1988. Studies on Atyi­dae (Decapoda, Caridea) of Sri Lanka. III.Aspects of the population ecology of Car­idina simoni Bouvier, 1904. Crustaceana(Leiden) 54:85-103.

Dudgeon, D. 1985. The population dynamicsof some freshwater carideans (Crustacea:Decapoda) in Hong Kong, with specialreference to Neocaridina serrata (Atyidae).Hydrobiologia 120:141-149.

Ellis-Neill, L. 1987. Distributional and pro­duction dynamics of benthic invertebratesin a tropical stream on Guam. M.S. thesis,University of Guam, Guam.

Fraser, D. F., J. F. Gilliam, and T Yip-Hoi.1995. Predation as an agent of populationfragmentation in a tropical watershed.Ecology 76:1461-1472.

Gilinsky, E. 1984. The role of fish predationand spatial heterogeneity in determiningbenthic community structure. Ecology65:455-468.

Hart, R C. 1981. Population dynamics andproduction of the tropical freshwatershrimp Caridina nilotica (Decapoda: Atyi­dae) in the littoral of Lake Sibaya. Fresh­water BioI. 11:531-547.

Helm, W. T, P. Brouha, M. Aceituna,C. Armour, P. Bisson, J. Hall, G. Holton,and M. Shaw. 1985. Glossary of streamhabitat terms. Western Division, AmericanFisheries Society.

Jalihal, D. R, G. B. Almelkar, and K. N.Sankolli. 1994. Atyid shrimps of the genusCaridina H. Milne Edwards, 1837: Poten­tial crustacean material for experimentalbiology. Crustaceana (Leiden) 66: 178­183.

Reice, S. R, and R L. Edwards. 1986. Theeffect of vertebrate predation on loticmacroinvertebrate communities in Que­bec, Canada. Can.J. Zoo1. 64:1930-1936.

Reznick, D. N., F. H. Shaw, F. H. Rodd, andR G. Shaw. 1997. Evaluation of the rate ofevolution in natural populations of guppies(Poecilia reticulata). Science (Washington,D.C.) 275:1934-1937.

Shokita, S. 1976. Early life-history of the

398

land-locked atyid shrimp, Caridina dentic­ulata ishigakiensis Fujino et Shokita, fromthe Ryukyu Islands. Res. Crustacea, Car­cino1. Soc. Jpn. 7:1-10.

Sokal, R. R., and F.]. Rohlf. 1995. Biometry:The principles and practice of statistics inbiological research, 3d ed. W. H. Freemanand Company, New York.

Wellborn, G. A., and]. V. Robinson. 1991.

PACIFIC SCIENCE· October 2001

The influence of fish predation in an ex­perienced prey community. Can. ]. Zoo1.69:2515-2522.

Wiles, G.]., and M. W. Ritter. 1993. Guam.Pages 129-178 in D. A. Scott, ed. A di­rectory of wetlands in Oceania. IWRB,Slimbridge, u.K., and AWE, Kuala Lum­pur, Malaysia.